Official-NV: An LLM-Generated News Video Dataset for Multimodal Fake News Detection

TL;DR

This paper introduces Official-NV, a new dataset of officially published news videos for multimodal fake news detection, and proposes OFNVD, a baseline model utilizing advanced attention mechanisms, demonstrating improved detection performance.

Contribution

The paper creates a high-quality, officially sourced news video dataset and proposes a novel multimodal detection model with a GLU attention mechanism and cross-modal Transformer.

Findings

The dataset effectively reduces noise compared to user-uploaded videos.

The OFNVD model outperforms existing baselines in fake news detection accuracy.

Benchmark results validate the model's effectiveness on the new dataset.

Abstract

News media, especially video news media, have penetrated into every aspect of daily life, which also brings the risk of fake news. Therefore, multimodal fake news detection has recently garnered increased attention. However, the existing datasets are comprised of user-uploaded videos and contain an excess amounts of superfluous data, which introduces noise into the model training process. To address this issue, we construct a dataset named Official-NV, comprising officially published news videos. The crawl officially published videos are augmented through the use of LLMs-based generation and manual verification, thereby expanding the dataset. We also propose a new baseline model called OFNVD, which captures key information from multimodal features through a GLU attention mechanism and performs feature enhancement and modal aggregation via a cross-modal Transformer. Benchmarking the dataset and baselines demonstrates the effectiveness of our model in multimodal news detection.